One record, prepared in a consistent structure, that each party in a placement can work from.
An AI risk rarely stays in one set of hands. It starts with the insured, moves to the broker, is placed with a carrier, and is often ceded in part to a reinsurer. At each hop someone has to understand what the applicant’s autonomous system does and whether the evidence behind the answers holds. Today that understanding is rebuilt at every step. The insured assembles answers for a questionnaire. The broker repackages them into a submission. The carrier re-verifies what it can and takes the rest on trust. If a claim comes, everyone reconstructs the story again, under pressure, from records that were never designed to be reused.
The problem is not that any one party is careless. It is that the evidence does not exist as a single record that travels. Each party works from its own version, and every translation loses fidelity and adds delay.
The chain runs on parties who do not fully trust each other
An insurance placement is a chain of parties with different interests: the insured wants cover, the broker wants to place it, the carrier wants to price it accurately, the reinsurer wants to understand what it is taking on. None of them can simply accept a version of the facts prepared by someone with a stake in the outcome. That structural distrust is why an insured’s own description of its AI governance carries limited weight, and why so much of a placement is spent re-establishing facts that were already stated once.
A record only crosses that boundary if it was prepared in a consistent structure rather than to taste. A figure in a set of audited accounts travels between a company, its bank and its regulator because it was prepared to a recognised structure, not because any of them trusts the company’s word. Evidence about an autonomous system needs the same property to move through a placement.
One record, prepared in a consistent structure, that each party can read
Decision provenance is a record of what an autonomous system did, on what basis, under whose authority, and how each item was captured, kept so it can be reconstructed later. Prepared in a consistent structure, it becomes an artefact each party in the chain can read the same way.
The insured’s submission, the broker’s placement file and the carrier’s underwriting file can draw on one record rather than three versions of a story. Because the record shows how each fact is known, whether it came from the infrastructure that ran the decision, was captured close to the event, was reconstructed, or was stated by a person, a party who did not prepare it can weigh it rather than take it on faith. That honesty about capture method is what lets the record survive the handoff. A record that presents everything as equally certain invites exactly the discounting that structural distrust produces.
Custara prepares, structures and preserves that record so it travels in a form each party can read. It does not price the risk, place or advise on cover, or reach an underwriting conclusion. What the record means at each stage is that party’s decision.
The record built at underwriting is the record that answers the claim
The portability that matters most shows up after a loss. When a claim turns on what an autonomous system actually did, the parties reconstruct the decision, and the quality of that reconstruction depends on what was captured at the time, not on what anyone can assemble afterward. A provenance record built at underwriting, and kept current through the policy period, is the same record that supports the reconstruction when a claim arrives. The evidence does not have to be manufactured under the worst conditions, after an outcome, with an interested party holding the pen.
A standard can define what the record should contain and how evidence quality is described, which is the work of the Decision Standards Institute, and an independent assessor can review the record separately, which is the role of Attestra. That separation is what lets a party who did not prepare the record review it without treating the preparer as the grader. Custara prepares the record. It does not assess it, price the risk, or decide a claim.
Why portability is the point
One record that the insured, the broker, the carrier and the reinsurer can each work from is not a convenience. It is the difference between a placement that spends its time re-establishing facts and one that spends its time pricing risk. It is also the difference between a claim reconstructed from a record built for the purpose and one assembled from logs that were not.
None of this promises any party will accept the record or reach a particular decision. Each party decides on its own terms. What a portable record changes is what they are deciding from: shared evidence, prepared in a consistent structure, honest about how each fact is known, rather than a different version of the story at every hop.
See how one record structures an insurer’s inbound AI review in the worked examples, or read how Custara works with insurers.
The framework behind this approach, and the book that sets it out, are forthcoming. This article is general information, not legal or insurance advice.